@Mark: Are your VMWare interface servers located on different physical hosts? Are they using the same switch? If not this would make a less robust failover solution as you single point of failure in your architecture.
With the PI OPC interface you can configure to have a primary and secondary interface which is ready to take over if a problem occurs. You can opt out for different strategies such as cold, warm and hot failover mode. You need to validate that the underlying infrastructure (PLC, SCADA, DCS) can support more network traffic as these different nodes can poll the same events multiple times.
I would suggest to read this manual to know the details about PI OPC DA Failover.